基于数字孪生的主动电网设备状态推演方法

A Digital Twin-Based Method for Equipment State Inference in Active Power Grids

主动电网的动态变化、网络异构和信息庞杂等特性给孪生体的高精准度映射带来了巨大挑战。为解决数字孪生主动电网的高精准度映射问题,提出了一种基于径向基函数神经网络的设备状态推演方法。方法通过建立径向基函数神经网络的电网设备状态监测器,获取正常负荷下的残差数据,并计算设备负荷系数。然后,利用基于马氏距离的相似矩阵识别设备的运行环境,并调用相应的设备推演模型,实现对复杂工况下设备状态的精确推演。实验证明,上述方法的设备状态推演准确率高达92%,满足数字孪生主动电网的要求,为电网智能调控和精确规划提供了基础信息,同时为电网数字化建设提供了理论参考。

The dynamic changes,network heterogeneity,and complex information in active power grids pose significant challenges for accurately mapping digital twins.To address the high precision mapping issue in digital twin active power grids,we proposed a device state inference method based on radial basis function neural network(RBFNN).This method involves establishing an RBFNN-based monitoring system for power grid devices,acquiring residual data under normal loads,and calculating device load coefficients.Subsequently,we used a Mahalanobis distance-based similarity matrix to identify the operating environment of devices and invoked the corresponding device inference model to achieve precise inference in complex operating conditions.Experimental results demonstrate an impressive device state inference accuracy of 92%,meeting the requirements for constructing digital twin active power grids.This method provides fundamental information for intelligent control and precise planning in power grids and serves as a theoretical reference for power grid digitalization.